Classic broker's desk with leather portfolio and fountain pen, digital data streams emerging — the old world meeting the new

The Broker Dies Last

Specification Standard: November 10, 2026 • Technical Protocol Spec v2026-11-10

Protocol Architecture Specification
COMMERCE & COORDINATION SPECIFICATION MATRIX: The Broker Dies Last (SPEC v2026-11-10)
Friction Component Legacy Broker Fee Structure True Underlying Compute Economic Rent Extracted UCP Protocol Cost Capital Efficiency Gain
Information Asymmetry Tax Percentage toll on deal value Minimal server compute Monopolizing private pricing data Open transparent feeds Complete price transparency
Transaction Escrow & Trust High payment intermediary fee Nominal bank interchange Excessive risk hedging markup AP2 atomic smart escrow Significant settlement fee reduction
Contract Generation & Paperwork Hourly legal documentation billing Minimal compute time Billing hours for boilerplate forms Standardized JSON-LD contracts Near-total legal friction elimination
Counterparty Risk Premium Arbitrary factored rate markup Statistical default risk Arbitrary risk assessment penalties Cryptographic collateral locking Instant liquidity access
Communication & Phone Tag Indirect labor delay cost Zero machine API latency Delay overhead across multiple intermediaries Sub-second agent negotiation Real-time deal finalization
Search & Discovery Markup Hefty referral commission cut Standard vector database query Controlling directory placement Open WebMCP discovery registry Zero customer acquisition tax

Will’s Take is editorial perspective — opinion, future-casting, and industry observation from Will Tygart. Not analysis. Not client work. Just how I see it.

Every time a new technology wave hits, someone writes the obituary for the middleman too early.

The internet was supposed to kill the insurance broker. The travel agent. The mortgage broker. The staffing agency. The real estate agent. Some of them got hurt. Most of them adapted. A few of them actually got stronger because the complexity of the underlying transaction didn’t go away — it just moved around.

Why agentic commerce is different

Agentic commerce is different. And I say that as someone who has watched a lot of “this changes everything” moments come and go.

The difference this time is that the agent isn’t just surfacing information faster. It’s executing. It’s not helping a buyer make a decision — it’s making the decision, completing the transaction, and reporting back. That’s a fundamentally different threat model for anyone whose value proposition is sitting in the middle of a transaction and adding judgment.

Let’s talk about what brokers actually do.

Not the theoretical version. The real version.

An insurance broker doesn’t just find you a policy. They translate your risk profile into carrier language. They know which carriers are hungry in which lines right now. They know which ones are pulling back from certain geographies or asset classes. They know how to structure a submission to get the best look. They know who to call when a claim goes sideways.

That’s not just information. That’s relationships, pattern recognition, and institutional memory built over years of reps.

A staffing broker doesn’t just match resumes to job descriptions. They know which hiring manager at which company is actually the decision maker versus the gatekeeper. They know when a client says they want X but they’ll actually hire Y. They know which candidates interview well but flame out and which ones undersell themselves. They know the unwritten rules of a dozen different industries.

A lending broker in the specialty asset space — watches, art, jewelry, classic cars — doesn’t just find capital. They know which lenders are over-allocated in certain categories. They know who will move fast and who will drag their feet. They know how to present an asset to get the best advance rate. They know the difference between a lender who will be a good partner if things get complicated and one who will be a problem.

None of that is in a database. None of that is queryable. None of that gets replaced by a capability profile.

Compression, not replacement

But here’s the thing.

The parts of brokerage that aren’t that — the administrative parts, the intake parts, the document collection parts, the status update parts, the scheduling parts — those are absolutely going to agents. And those parts currently consume a significant percentage of what brokers spend their time on.

That’s the real threat. Not replacement. Compression.

When an agent can handle first contact, intake, document collection, preliminary underwriting data assembly, and status communication — the broker is left with the pure judgment work. Which sounds great until you realize that most brokerage businesses are priced and staffed assuming the administrative overhead is part of the package.

The economics change. The headcount model changes. The revenue per broker goes up but the total broker count goes down. The ones who survive are the ones who are genuinely, irreplaceably good at the judgment part.

The broker dies last

The broker dies last because they’re the hardest to fully replace.

Not because they’re protected. Because the underlying transaction is genuinely complex and the liability for getting it wrong falls on a human who has to sign something.

Insurance placements have E&O exposure. Lending transactions have regulatory frameworks. Staffing placements carry co-employment risk. These aren’t just commerce transactions — they’re relationships with legal and financial consequence on both sides.

Agents will handle more and more of the workflow. But someone still has to own the outcome.

For now.

The judgment layer

The brokers who understand what’s happening and start positioning themselves as the judgment layer — the ones who explicitly offload the administrative overhead to agents and double down on the relationship and expertise work — those are the ones who make it to the other side of this transition.

The ones who keep running their business the same way, treating every status update call and document request as a billable relationship touchpoint, are going to find that clients stop paying for what an agent will do for free.

I work with enough of these businesses to see both sides of this.

The operators who are genuinely excellent — who have real expertise, real relationships, real pattern recognition — aren’t scared of this. They’re relieved. The administrative grind is the part they hate. If agents take that, they get to do more of what they’re actually good at.

The operators who are running on volume and relationships and not much else underneath — they feel it differently. Because the volume part is exactly what agents eat first.

The broker dies last. But some brokers die sooner than others.

More from Will’s Take

  • I Want an Agent That Negotiates My Watch Service
  • What Happens to Price When Every Buyer Has an Agent?
  • I Don’t Think UCP Is About Shopping
  • What UCP Means for the Guy Who Does One Thing Really Well
  • The Insurance Carrier Is the Most Important UCP Node Nobody’s Talking About

Related Deep Dives

  • Agent-to-Agent Commerce: AI Systems Negotiating Deals
  • Agent Vendor Lock-in: Switching Costs in Agentic Commerce
  • Agent Commerce Margin Compression

Frequently Asked Questions: Sovereign Systems & Market Coordination

Q1: What economic definition explains why brokers historically captured such high margins?

Brokers exploited information asymmetry and trust deficits: parties lacked verified pricing data and trusted clearing mechanisms, forcing them to pay high tolls to access markets.

Q2: How do open protocols break the cycle of new platforms simply replacing old brokers?

Closed platforms become the new super-brokers; open protocols (like UCP) are non-profit public standards where no single entity can hike take rates.

Q3: Why did previous attempts at disintermediation in the 1990s and 2000s fail?

Early web standards lacked native cryptographic identity, standardized data contracts, and automated financial clearing, forcing markets back into the hands of centralized platforms.


Posted

in

by

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *